SOX1 Promotes Differentiation of Nasopharyngeal Carcinoma Cells by Activating Retinoid Metabolic Pathway

Lei et al. Cell Death and Disease (2020) 11:331 https://doi.org/10.1038/s41419-020-2513-1 Cell Death & Disease

ARTICLE Open Access SOX1 promotes differentiation of nasopharyngeal carcinoma cells by activating retinoid metabolic pathway Xin-Xing Lei1,YunLiu2, Jin-Xing Wang3,QianCai2,MinYan1,Hui-PingHe1,QuentinLiu1,3,Zi-JieLong3 and Zhong Guan2

Abstract Undifferentiation is a key feature of nasopharyngeal carcinoma (NPC), which presents as a unique opportunity for intervention by differentiation therapy. In this study, we found that SOX1 inhibited proliferation, promoted differentiation, and induced senescence of NPC cells, which depended on its transcriptional function. RNA-Seq- proﬁling analysis showed that multiple undifferentiated markers of keratin family, including KRT5, KRT13, and KRT19, were reduced in SOX1 overexpressed NPC cells. Interestingly, gene ontology (GO) analysis revealed genes in SOX1 overexpressed cells were enriched in extracellular functions. The data of LC/MS untargeted metabolomics showed that the content of retinoids in SOX1 overexpressed cells and culture medium was both higher than that in the control group. Subsequently, we screened mRNA level of genes in retinoic acid (RA) signaling or metabolic pathway and found that the expression of UDP-glucuronosyltransferases was signiﬁcantly decreased. Furtherly, UGT2B7 could rescue the differentiation induced by SOX1 overexpression. Inhibition of UGTs by demethylzeylasteral (T-96) could mimic SOX1 to promote the differentiation of NPC cells. Thus, we described a mechanism by which SOX1 regulated

1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; the differentiation of NPC cells by activating retinoid metabolic pathway, providing a potential target for differentiation therapy of NPC.

Introduction offer new opportunities for treating NPC. Recently, Nasopharyngeal carcinoma (NPC) is endemic in remarkable achievements have been made by differ- South-Eastern Asia, and northern Africa. At present, the entiation therapy in the treatment of some types of main treatment approaches for NPC include radio- tumors3,4. 95% of NPC is pathologically undifferentiated therapy, chemotherapy, and targeted therapy. Radiation squamous cell carcinoma, which make NPC an appro- therapy is the firstchoiceforthetreatmentofNPC. priate model for differentiation therapy. However, it is inevitable that nasopharyngeal mucosa SOX1 encodes a member of the SOX (SRY-related and tissue will be damaged after radiotherapy1,2. Thus, high mobility group box) family of transcription factors more effective and less toxic regimens are needed to involved in the regulation of embryonic development – and determination of the cell fate5 7.SOX1shares similar biochemical properties and functions with SOX2 Correspondence: Quentin Liu ([email protected])or or other members, but the different affinity to target Zi-Jie Long ([email protected]) or Zhong Guan ([email protected]) genes, post-transcriptional modifications and interac- 1Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in South China, 510060 Guangzhou, China tions with other co-factors lead to different functions in – 2Department of Otorhinolaryngology, Sun Yat-sen Memorial Hospital, Sun Yat- distinct biological contexts8 10. SOX proteins bind sen University, 510120 Guangzhou, China sequence-specifically to DNA by a high-mobility group Full list of author information is available at the end of the article These authors contributed equally: Xin-Xing Lei, Yun Liu (HMG) domain that allows them to function as Edited by A. Stephanou

© The Author(s) 2020 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a linktotheCreativeCommons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/.

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 2 of 15

transcription factors8,11. Gene silencing in cancer is V5-tagged SOX1, ΔHMG SOX1, or Δ246-391 SOX1. The associated with promoter hypermethylation12 and SOX1 primers used for gene cloning and plasmid construction hypermethylation was detected in primary tumor tis- are listed in Supplementary Table S1. sues13. SOX1 inhibits cell proliferation, reduces migration and invasion ability, and induces differentiation of Lentiviral production, infections, and cell line generation – tumor cells14 17. Recently, we found that SOX1 was Lentivirus was produced by transient transfection by associated with decreased expression of undifferentiation Lipofectamine 2000 (Invitrogen) in human embryonic markers (KRT13 and KRT19) in NPC cells16. However, kidney (HEK) 293T cells using with a second-generation the mechanism by which SOX1 promotes NPC cell dif- lentiviral vector system. All virus-containing medium was ferentiation has not been established. mixed with 8 μg/mL polybrene (Sigma-Aldrich). Virus Retinoic acid (RA), especially all-trans-retinoic acid produced from pLVX-Tet3G plasmid was used to infect (ATRA), is the most potent natural metabolite of vitamin wild type HONE1 or CNE2 cells to construct HONE1- A. RA is involved in a variety of biological functions Tet-On or CNE2-Tet-On stable cell lines. Cells were including embryogenesis, cell differentiation, and apop- selected in 1 mg/mL G418 for at least 2 weeks. Subse- tosis18,19. Retinoids induce differentiation and/or apop- quently, HONE1-Tet-On and CNE2-Tet-On cell lines tosis in tumor cells and show anti-proliferative and anti- were infected by virus produced from pLVX-TRE-(X) oxidant activity, which have great potential as therapeutic plasmids and selected by 2 μg/mL puromycin for 6 days. or preventive agents. The distribution and level of RA in HONE1TRE-SOX1 and CNE2TRE-SOX1 cells infected by virus embryonic tissues are tightly controlled by oxidation produced from pLVX-UGT1A6 or pLVX-UGT2B7 plas- through retinol and retinal dehydrogenases and by mids were selected by 12 μg/mL blasticdin for 7 days. hydroxylation via specific cytochrome P450 family 26 enzymes (CYP26s)20,21. Cell culture and doxycycline induction UDP-glucuronosyltransferases (UGTs) are phase II The HONE1 and CNE2 cell lines were obtained from metabolism isoenzymes that are found to be active in Dr.Chao-NanQian(SunYat-senUniversity,Guangz- liver, kidneys, epithelial cells of the lower gastrointestinal hou, China). Wild type cell lines and their lentiviral- tract and brain20,22. Although the characterizations of infected stable cell lines were all maintained in RPMI UGT gene family members were identified, their roles in 1640 (Invitrogen) supplemented with 10% fetal bovine clearance and homeostasis of endogenous substrates are serum (FBS, Hyclone). The cells were incubated at 37 °C still insufficiently understood. Previous study proves that in a humidified chamber containing 5% CO2.HONE1- retinoids and their analogs are glucuronidated by Tet-On-(X) and CNE2-Tet-On-(X) cell lines were trea- UGTs20,23,24. Excess retinoids can also inhibit the ted with 2 μg/mL doxycycline (Medchem Express) to expression of UGTs, indicating the existence of regulatory induce overexpression of SOX1 or its mutant/ loops and complexity25. truncated form. In the present study, we showed that overexpression of SOX1 promoted NPC cells to differentiate via its tran- Cell viability assay scriptional function. These new data also established a Cells were seeded onto 96-well plates at an initial mechanism that retinoids were accumulated by SOX1, density of 2 × 103 cells/well. At specified time points, contributing to the differentiation of NPC cells, which 10 μL of CCK-8 solution was added to each well of the offered a novel approach for NPC differentiation therapy. plate. Then the plate was incubated for another 2 h. Cell viability was determined by scanning with a microplate Materials and methods reader at 450 nm. Plasmid constructs The plasmids encoding human SOX1 (wild type, WT), Immunofluorescence staining mutant SOX1 (SOX1mut_2), V5-tagged SOX1, ΔHMG Cells were fixed in 4% paraformaldehyde at room SOX1, and Δ246-391 SOX1 were generated by PCR temperature for 10 min and permeabilized in 0.5% Tri- amplification and subcloned into the pLVX-TRE3G ton X-100 in PBS for 10 min. Slides were incubated with expression vector; UGT1A6 and UGT2B7 were sub- the primary antibody (1:200 dilution) overnight. Immune cloned into pLVX expression vector; promoters of complexes were stained with the secondary antibody UGT1A6 and UGT2B7 were subcloned into pGL3-basic conjugated to Alexa-488 or Alexa-546 (Invitrogen, 1:200 vector using the ClonExpress II One Step Cloning Kit or dilution) at room temperature for 1 h. Nuclei were ClonExpress MultiS One Step Cloning Kit (Vazyme) stained with DAPI (Sigma-Aldrich) and viewed with an according to the manufacturer’s instructions. The new Olympus IX71 microscope. The following antibodies plasmid was named pLVX-TRE-(X), where X stood were used as primary antibodies: E-cadherin and V5-tag for SOX1 (wild type, WT), mutant SOX1 (SOX1mut_2), (Cell Signaling Technology 3195 and 13202,

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 3 of 15

respectively), Ki-67 (Proteintech 27309-1-AP), and Gene set enrichment analysis (GSEA) SOX1 (GeneTex EPR4766). GSEA was performed using hallmark gene sets from the Molecular Signatures Database (MSigDB). RNA-Seq Colony formation assay expression data sets were loaded into GSEA 3.0 software. Approximately 500 cells were seeded into six-well plates 10,000 permutations were done and finally a list of gene set in triplicate and incubated for 8 days. Colonies were ranks with information of normalized enrichment scores stained with crystal violet and viewed. (NES), P-value and FDR-q-value was obtained. The top significant gene sets were viewed and sorted. Senescence-associated β-galactosidase (SA-β-gal) staining Cultured cells were washed in PBS and SA-β-gal Conditional medium (CM) preparation activity was detected using senescence β-galactosidase For collecting CM of NPC cells, HONE1-Tet-On staining kit (Beyotime) according to the manufacturer’s SOX1 cells were cultured in RPMI 1640 supplemented direction. with 10% FBS. At a 50% confluence, medium was dis- carded and cells were incubated with fresh RPMI 1640 Cell lysis and western blot analysis (10% FBS) with or without 2 μg/mL doxycycline for 48 h. Cells were lysed on ice in RIPA buffer. Protein con- The CM was then collected, filtered using a 0.22 μm filter, centration was determined by using the Bradford dye and stored at −80 °C until use. method. Equal amounts of cell extracts were subjected to electrophoresis in 10% gradient SDS–PAGE gels and Metabolic profiling by LC–MS then transferred to 0.45 µm PVDF membranes (Milli- LC–MS analysis was performed on an ACQUITY pore) for antibody blotting. The following antibodies UHPLC system (Waters Corporation) coupled with an AB were used: V5-tag, p21 Waf1/Cip1, Rb, Phospho-Rb SCIEX Triple TOF 5600 System (AB SCIEX) equipped (Ser780), Phospho-mTOR (Ser2448), Phospho-p70 S6 with an electrospray ionization source, at both positive Kinase (Thr389), Phospho-p70 S6 Kinase (Ser371), and and negative ion modes. Each group contains five repli- Phospho-4E-BP1 (Thr37/46) (Cell Signaling Technology cative samples. The stability of LC–MS analytical systems 13202, 2947, 9309, 9307, 2971, 9205, 9208 and 2855, was evaluated using the pooled QC samples at regular respectively), SOX1 (GeneTex EPR4766), KRT19 intervals (every five samples). The LC–MS data were (Abcam EP1580Y), KRT13, mTOR, p70 S6 Kinase and exported as .wiff files and converted to .mzXML files 4E-BP1, (Epitomics 2713-1, 1612-1, 1494-1, and 1557-1, using msconvert software (ProteoWizard tool). The files respectively), KRT5, CDK4, CDK6, c-Myc, PPARγ,and were then uploaded on XCMS database (https:// β-actin (Proteintech 66727-1-Ig, 11026-1-AP, 14052-1- xcmsonline.scripps.edu) for further analysis according to AP, 10828-1-AP, 16643-1-AP, and 60008-1-Ig, respec- XCMS instruction. tively), UGT1A6 and UGT2B7 (Signalway Antibody 43176 and C32390, respectively). Horseradish Cell cycle, apoptosis, and dual-luciferase reporter assay peroxidase-conjugatedgoatanti-mouseorgoatanti- Cells were collected and fixed in 70% ethanol at −20 °C rabbit IgG (Pierce) was used as a secondary antibody. overnight and then washed and resuspended in FxCycle™ Proteins were visualized with Immobilon Western Che- PI/RNase Staining Solution (Invitrogen). After incubating milum HRP Substrate (Millipore). in the dark for 30 min at room temperature, cell cycles were analyzed by flow cytometry. RNA-Seq Cells were collected and resuspended in the binding RNA-Seq data generation and normalization were per- buffer and then Annexin-V-FITC and PI (Invitrogen) formed on an Illumina HiSeq™ PE150 system by the were added to the cells according to the protocol. The Novogene Bioinformatics Technology Co., Ltd. (Beijing, cells were then incubated for 15 min in the dark and China). subjected to flow cytometry for apoptosis analysis. Luciferase activity in cell lysates was determined by the Quantitative real-time PCR (RT-PCR) dual-luciferase reporter assay system (Promega). Firefly Total RNA was extracted by using HiPure Total RNA luciferase data for each sample were normalized against Kits (Magen), which was used to generate cDNA by Renilla luciferase activity. using One-Step RT-PCR SuperMix (TransScript). Quantitative RT-PCR was performed using ChamQ Statistical analysis SYBR qPCR Master Mix (Vazyme) as recommended by Each experiment was performed in triplicate and the manufacturer. The primers used were listed in repeated at least three times. Unless otherwise indicated, Supplementary Tables S2–4. ACTB was used as the data were presented as mean ± SD of three independent internal control. experiments. Statistics were calculated by Prism

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 4 of 15

Fig. 1 (See legend on next page.)

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 5 of 15

(see figure on previous page) Fig. 1 SOX1 overexpression promotes differentiation of NPC cells. a Morphology of HONE1TRE-SOX1 and CNE2TRE-SOX1 cells with doxycycline treatment for 72 h. Scale bar = 50 µm. b Confocal immunofluorescence for E-cadherin (green) and DAPI (blue) in HONE1TRE-SOX1 and CNE2TRE-SOX1 cells with or without doxycycline treatment for 96 h. Scale bar = 30 μm. c Cell viability of HONE1TRE-SOX1 and CNE2TRE-SOX1 cells with (red) or without (blue) doxycycline treatment by CCK-8 assay. All data represent the mean ± SD (n = 4, **P < 0.01, ***P < 0.001, ****P < 0.0001). d Colony formation assay of HONE1TRE-SOX1 and CNE2TRE-SOX1 cells with or without doxycycline treatment for 8 days. Dot plots show number and average size of colonies calculated by imageJ software. All data represent the mean ± SD (n = 3, **P < 0.01, ***P < 0.001). e Confocal immunofluorescence (left panel) for Ki-67 (red) and DAPI (blue) in HONE1TRE-SOX1 and CNE2TRE-SOX1 cells with or without doxycycline treatment. Scale bar = 50 μm. Dot plots (right panel) show quantification of the frequency of Ki-67-positive cells in each vision. All data represent the mean ± SD (n = 10, ****P < 0.0001). f SA-β gal staining (left panel) of HONE1TRE-SOX1 and CNE2TRE-SOX1 cells with or without doxycycline treatment. Red arrows represent SA-β gal-positive cells. Scale bar = 50 μm. Dot plots (right panel) show quantification of the frequency of SA-β gal-positive cells in each vision. All data represent the mean ± SD (n = 10, ****P < 0.0001).

Graphpad software (version 7.0). Differences among namely ΔHMG SOX1 (SOX1 without the N-terminus) variables were assessed by two-tailed Student’s t-tests and Δ246-391 SOX1 (SOX1 without the C-terminus), and one-way or two-way ANOVA with multiple com- together with a full-length SOX1 expression vector (V5- parisons tests. A P-value < 0.05 was considered statisti- tagged SOX1) (Supplementary Fig. 2a). All of the vec- cally significant (*P <0.05,**P < 0.01, ***P <0.001,****P tors were transferred to HONE1 and CNE2, respec- < 0.0001). tively. The cellular morphology and related differentiation protein makers of two truncated SOX1 Results expressed NPC cells were similar to the vehicle group Overexpression of SOX1 promotes the differentiation of (empty vector) (Supplementary Fig. 2b–d). Moreover, NPC cells two truncated SOX1 expression did not induce cell To determine whether SOX1 was required for the senescence (Supplementary Fig. 2e). These data showed development of differentiated NPC status, stable clones that loss of the HMG box and the C-terminus (tran- overexpressing SOX1 were selected from two poorly dif- scription activation domain) of SOX1 significantly ferentiated NPC cell lines (HONE1 and CNE2) using the reduced the ability of SOX1 to promote NPC cell dif- Tet-ON system. Expression of SOX1 was induced by ferentiation. HMG box is essential for binding with doxycycline compared to the cells without doxycycline. DNA. The homology of HMG box of SOX1 was highly Induction of SOX1 resulted in morphologic changes similar with other SOX subfamily members, especially histologically characterized as slender and fusiform phe- SOX2 (Supplementary Fig. 3). Thus, we mutated two notype in HONE1 and CNE2 cells (Fig. 1a). E-cadherin amino acids (Arg53 to Asp and Asn78 to Ala) in the was up-regulated in HONE1 and CNE2 cells with SOX1 DNA-binding sites of SOX1 according to SOX2 crystal overexpression (Fig. 1b). In addition, overexpression of structure(SupplementaryFig.4,Fig.2b). The expres- SOX1 attenuated cell proliferation and colony formation sion of mutant SOX1 was still localized in the nucleus as (Fig. 1c–e, Supplementary Fig. 1). Besides, the proportion detected by immunofluorescence staining (Fig. 2d). The of senescence-associated β-gal-positive cells was cellular morphology and related differentiation protein increased following SOX1 overexpression (Fig. 1f). The makers of the mutant SOX1 expressed NPC cells were results indicated that SOX1 promoted the differentiation similar to the vehicle group (empty vector) (Fig. 2c, e). of NPC cells. Moreover, mutant SOX1 expression did not induce cell senescence (Fig. 2f). The data showed that mutant of the SOX1 promoting NPC cell differentiation depends on its HMG box in SOX1 significantly impaired the ability of transcriptional function SOX1 to promote NPC cell differentiation. Taken The mechanism underlying the role of SOX1 during together, these findings indicated that SOX1 induced the cellular differentiation process of NPC remained NPC cell differentiation attributing to its transcriptional largely elusive. Previous study showed that SOX family function. members regulated developmental events not only by acting as direct transcriptional regulators, but also by SOX1 induces a global alteration of keratins forming protein–protein interactions and acting as Although it has been reported that keratins were used either coactivators or co-suppressors8. To reveal the as indicators of differentiation, there was no systematic underlying mechanism, we firstly detected SOX1 studies on how the family of keratins altered between expression in HONE1 and CNE2 cells and found that differentiated and undifferentiated NPC. Here, we ana- SOX1 was localized in the nucleus (Fig. 2a). Then, two lyzed RNA-Seq data of 42 cases of carcinoma tissues as truncated SOX1 expression vectors were also made, well as 4 cases of normal tissues from GEO database. A

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 6 of 15

Fig. 2 HMG box is indispensable for SOX1 to promote NPC cell differentiation. a Confocal immunofluorescence for SOX1 (red) and DAPI (blue) in HONE1 and CNE2 cells. Fluorescence images are respectively merged to display location of SOX1 (red). Scale bar = 4 µm. b Schematics of full length and mutated SOX1 proteins used in this study. c Morphology of HONETRE-(X) and CNE2TRE-(X) cells (X stand for vehicle, wild type SOX1 or mutant SOX1) under doxycycline treatment for 3 days. Scale bar = 50 µm. d Confocal immunofluorescence for SOX1 (red) and DAPI (blue) in HONETRE-(X) and CNE2TRE-(X) cells under doxycycline treatment for 3 days. Scale bar = 4 μm. e Western blot analysis of SOX1, KRT5, KRT13, and β-actin expression in HONETRE-(X) and CNE2TRE-(X) SOX1 cells under doxycycline treatment for 3 days. β-actin was used as a control. f SA-β gal staining (left panel) of HONETRE-(X) and CNE2TRE-(X) cells under doxycycline treatment for 7 days. Red arrows represent SA-β gal-positive cells. Scale bar = 50 μm. Dot plots (right panel) show quantification of the frequency of SA-β gal-positive cells in each vision. All data represent the mean ± SD (n = 5, ****P < 0.0001).

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 7 of 15

Fig. 3 Screening of NPC cells undifferentiated markers in keratin gene family. a RNA-Seq analysis displays heat map of keratins gene family expressed in 42 Chinese NPC patients and 4 non-NPC tissues from GEO database (GSE68799). mRNA intensities were rlog transformed and are displayed as colors ranging from red to blue. Both rows and columns are clustered using correlation distance and average linkage. b RNA-Seq analysis shows heat map of keratin gene family expressed in HONE1TRE-SOX1 (left) and CNE2TRE-SOX1 (right) cells under doxycycline treatment for 4 days. mRNA intensities were rlog transformed and are displayed as colors ranging from red to blue. Both rows and columns are clustered using correlation distance and average linkage. c RT-PCR analysis of keratin genes in HONE1TRE-SOX1 (upper) and CNE2TRE-SOX1 (lower) cells with (red) or without (blue) doxycycline treatment for 48 h. Data are normalized by the amount of ACTB mRNA and represent mean ± s.e.m. (n = 3, n.s.: P > 0.05, ***P < 0.001, ****P < 0.0001). d Western blot analysis of keratin proteins, SOX1, and β-actin in HONE1TRE-SOX1 and CNE2TRE-SOX1 cells with or without doxycycline treatment for 96 h. β-actin was used as a loading control. s.e.: short exposure, l.e.: long exposure. e GSEA of SOX1 (Dox+)vs.control(Dox−)inHONE1TRE-SOX1 and CNE2TRE-SOX1 cells using ‘Hallmark G2M Checkpoint’, ‘Hallmark E2F Targets’,and‘Hallmark mitotic spindle’ gene sets. NES normalized enrichment score, FDR false discovery rate.

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 8 of 15

Fig. 4 (See legend on next page.)

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 9 of 15

(see ﬁgure on previous page) Fig. 4 Untargeted LC/MS-based metabolomics reveals enrichment of retinoid pathway in SOX1 overexpressed cells. a Signiﬁcantly upregulated (upper) or down-regulated (lower) mRNAs (|fold-change | ≥2andP < 0.05) in HONE1TRE-SOX1/CNE2TRE-SOX1 (Dox+)comparedto corresponding control groups (Dox−) were shown by venn diagrams. b GO analysis of the differently expressed genes. Blue column represents

the total number of genes annotated with each GO term, while red is −log10 of P value. c Western blot analysis of keratin proteins and β-actin of wild type HONE1 cultured with conditional-media (CM) of HONE1TRE-SOX1 cell with (SOX1) or without (vec) doxycycline treatment for 48 h. β-actin was used as a loading control. d Differential feature plot for CM and cells of HONE1TRE-SOX1 with or without doxycycline treatment by LC–MS untargeted metabolomics. Only features that are dysregulated (P-value ≤ 0.05, fold change ≥ 1.5) are displayed. Upregulated features are shown in green, while downregulated features in red. The size of each bubble corresponds to the log fold change of that feature. The shade of the bubbles corresponds to the magnitude of the P-value (the darker the color, the smaller the P-value). Red arrows represent metabolites in retinoid pathway. e Summary of fold change, P-value, mass-to-charge ratio (m/z), and retention time (rt) of metabolites in retinoid pathway screened in d. f Western blot analysis of KRT5, KRT13, and β-actin of wild type HONE1 and CNE2 cells with or without RAce treatment for 72 h. β-actin was used as a loading control. g Colony formation assay of wild type HONE1 and CNE2 cells with vehicle, RA (10 μM), or RAce (10 μM) treatment for 8 days. h Cell viability of wild type HONE1 and CNE2 cells with (red) or without (blue) doxycycline treatment by CCK-8 assay. All data represent the mean ± SD (n = 4, ****P < 0.0001).

global high expression of keratins in NPC tissues was undifferentiated cells (Fig. 4b). To verify our hypothesis, found (Fig. 3a). In order to understand whether the we treated HONE1 cells with CM obtained from expression of SOX1 in NPC cells could influence the HONE1TRE-SOX1 cell culture medium (Dox+). Surpris- expression of keratins, we treated HONE1TRE-SOX1 and ingly, CM-induced alteration of KRT13 and KRT19 (Fig. CNE2TRE-SOX1 with or without doxycycline for 4 days, 4c). Then LC–MS untargeted metabolomics was used to and then RNA-Seq was performed. It showed that the analyze the composition of the CM and cells of induction of SOX1 greatly reduced the expression of HONE1TRE-SOX1 with or without doxycycline treatment. keratins (Fig. 3b). Moreover, RT-PCR and western blot The results showed that the content of retinoids increased assay were performed to verify the expression patterns of in both CM and cells with doxycycline treatment (Fig. 4d, keratins at mRNA and protein level (Fig. 3c, d). The e). To determine whether cell differentiation was due to results showed SOX1 decreased the expression of kera- elevated retinoids level, NPC cells were treated with tins, including KRT5, KRT13, and KRT19, which were retinol acetate (RAce) and subjected to detection of pro- basically consistent with the results of RNA-Seq data. tein levels. The expression of KRT5 and KRT13 was Thus, these data suggested that low expression of SOX1 decreased significantly, which demonstrated that retinoid was associated with high expression of keratins in NPC. metabolic pathway was involved in NPC cellular differ- GSEA displayed that genes were enriched in regulation entiation (Fig. 4f). Moreover, RA/RAce remarkably of G2/M checkpoint, E2F targets and mitotic spindle inhibited colony formation and cell viability of NPC cells after SOX1 overexpression (Fig. 3e). Moreover, as cell (Fig. 4g, h). These data suggested that SOX1 promoted differentiation usually require G0/G1 arrest, and E2F differentiation of NPC cells by activating retinoid targets were enriched after SOX1 overexpression, cell pathway. cycle distribution was then detected. Results showed that SOX1 induced cell cycle G0/G1 phase arrest (Supple- UGT2B7 disrupts SOX1 to promote differentiation of NPC mentary Fig. 5a, b), but did not promote obvious apop- cells tosis (Supplementary Fig. 6). GSEA analysis revealed that Our data showed that the content of retinoids was SOX1 downregulated Myc target genes and inhibited increased in differentiated NPC cells due to over- mTOR1 pathway. SOX1 could significantly decrease the expressed SOX1. Retinoids signaling and metabolism expression of c-Myc and mTOR1 downstream proteins diagrams were drawn to represent how retinol transports in NPC cells (Supplementary Fig. 5c–f). to cells and converts to RA (Fig. 5a, c). The content of RA in cells is tightly controlled by numerous enzymes SOX1 promotes differentiation of NPC cells by activating involved in retinoid metabolism. Thus, the mechanism retinoid pathway of SOX1 increasing RA accumulation in NPC cells was RNA-Seq displayed significantly 229 up-regulated genes investigated. RT-PCR was performed to detect the and 95 down-regulated genes in HONE1TRE-SOX1 and expression of RA signaling pathway-related enzymes or CNE2TRE-SOX1 (Dox+) compared to corresponding con- receptors: the RA-inducible gene stimulated by retinoic trol groups (Dox−) (Fig. 4a). Gene ontology (GO) analysis acid 6 (STRA6), cellular retinoic acid-binding protein 1 of differentiated cells showed that genes were enriched in (CRABP1), cellular retinoic acid-binding protein 2 extracellular activities, suggesting that the extracellular (CRABP2), RARs (RARA, RARB, and RARG) and RXRs components of differentiated cells might be distinct from (RXRA, RXRB, and RXRG). Moreover, lecithin retinol

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 10 of 15

Fig. 5 (See legend on next page.)

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 11 of 15

(see figure on previous page) Fig. 5 SOX1 deregulates UGTs expression to activate retinoid pathway in NPC cells. a A brief overview of retinoic acid signaling pathway. Retinol transports to cells in a complex with CRBP through vitamin A receptor STRA6. In cytoplasm, retinol is oxidized and converted to RA. RA can complex with CRABP1/2 and transports to the nucleus. Following, RA forms a complex with RXR–RAR or RXR–PPARβ/δ heterodimers and binds to DNA of retinoic acid response element (RARE) or PPAR response element (PPRE), thereby activating transcription of target genes. b RT-PCR analysis of STAR6, CRABP1, CRABP2, RARA, RARB, RARG, RXRA, RXRB, and RXRG genes in HONE1TRE-SOX1 (left) and CNE2TRE-SOX1 (right) cells with (red) or without (blue) doxycycline treatment for 48 h. Data are normalized by the amount of ACTB mRNA, and compared to the corresponding value for cells without doxycycline treatment. Data represent mean ± s.e.m. (n = 3, n.s.: P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). c A brief overview of all- trans-retinoic acid metabolic pathway. In cells, all-trans-retinol can be converted to all-trans-retinal by alcohol dehydrogenase (ADH) or short-chain dehydrogenase/reductase (SDR), or to all-trans-retinyl esters by lecithin retinol acyltransferase (LRAT). All-trans-retinal can be further converted to all- trans-retinoic acid by aldehyde dehydrogenase (ALDH). Finally, all-trans-retinoic acid is metabolized to inactive retinoids by CYP26s or UGTs. d RT-PCR analysis of LRAT, CYP26A1, CYP26B1, CYP26C1, UGT1A (total), UGT1A1, UGT1A6, UGT1A9, UGT2B7, and UGT8 genes in HONE1TRE-SOX1 (left) and CNE2TRE-SOX1 (right) cells with (red) or without (blue) doxycycline treatment for 48 h. Data are normalized by the amount of ACTB mRNA and represent mean ± s.e.m. (n = 3, n.s.: P > 0.05, *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001). e Western blot analysis of SOX1, UGT2B7, KRT5, KRT13, and β-actin expression in HONE1TRE-SOX1, CNE2TRE-SOX1 as well as UGT2B7 overexpressed HONE1TRE-SOX1 and CNE2TRE-SOX1 cells under doxycycline treatment for 4 days. β-actin was used as a control. f Cell viability of HONE1TRE-SOX1, CNE2TRE-SOX1 as well as UGT2B7 overexpressed HONE1TRE-SOX1 and CNE2TRE-SOX1 cells with or without doxycycline treatment by CCK-8 assay. All data represent the mean ± SD (n = 4, ****P < 0.0001). g SA-β gal staining (left panel) of HONE1TRE-SOX1 and UGT2B7 overexpressed HONE1TRE-SOX1 cells under doxycycline treatment for 7 days. Red arrows represent SA-β gal- positive cells. Scale bar = 50 μm. Dot plots (right panel) show quantification of the frequency of SA-β gal-positive cells in each vision. All data represent the mean ± SD (n = 5, n.s.: P > 0.05, ****P < 0.0001). acyltransferase (LRAT), cytochrome P450 family therapy in NPC by revealing several new findings: (1) 26 subfamily (CYP26A1, CYP26B1, and CYP26C1), and SOX1 was involved in promoting differentiation of NPC UDP glucuronosyltransferase family (UGT1A (total), cells depending on its transcriptional function. (2) Low UGT1A1, UGT1A6, UGT1A9, UGT2B7, and UGT8) expression of SOX1 was associated with high expression genes were also detected (Fig. 5b, d). The data showed of keratins, similar to clinical NPC tissues. (3) SOX1 that SOX1 suppressed several UGT genes expression, promoted differentiation of NPC cells through increasing including UGT1A6 and UGT2B7 (Fig. 5d). Then dual- retinoids accumulation within or outside NPC cells, luciferase reporter assay revealed that SOX1 did not which was associated with decreasing expression of affect UGT1A6 or UGT2B7 promoters’ transcriptional UGTs. (4) T-96, a small molecular inhibitor of UGT1A6 activity (Supplementary Fig. 7). We continued to over- and UGT2B7, induced cell differentiation and could be express UGT1A6 or UGT2B7 in SOX1 ectopic expressed used in differentiation therapy for NPC. cells, and found that UGT2B7, but not UGT1A6, could In the present treatment of NPC, radiotherapy displays partially rescue the ability of SOX1 to induce NPC cell favorable therapeutic outcomes for early stage patients. differentiation (Fig. 5e–g, Supplementary Fig. 8). These For locoregionally advanced NPC patients, platinum- data indicated that UGT2B7 could be the target of SOX1. based concurrent chemoradiotherapy is the standard However, RA metabolic network regulated by SOX1 was therapeutic regimen. Induction chemotherapy together coordinately balanced by multiple factors, but not only with chemoradiotherapy could significantly improve UGT2B7. progression-free survival and overall survival compared UGTs glucuronidate RA and disrupt its biological with chemoradiotherapy alone2. However, local failure activity. SOX1 might prevent the glucuronidation of and distant metastasis are still the major challenges for RA by reducing the expression of UGTs and eventually the poor survival of advanced NPC patients, and other lead to the accumulation of RA in cells to promote therapeutics are needed to be developed for better differentiation. Then, HONE1 and CNE2 cells were outcome. treated with T-96, an inhibitor of UGT1A6 and Differentiation therapy has achieved promising results UGT2B7, and the cellular morphology was trans- in certain types of cancers, especially in leukemia3,4.More formed into differentiated status (Fig. 6a). Western blot than 95% NPC are type III undifferentiated carcinoma. also revealed that the relevant differentiation markers, Patients diagnosed with undifferentiated carcinoma have KRT5 and KRT13, were all decreased (Fig. 6b). T-96 a poor prognosis and high mortality rate, which calls for could also inhibit cell proliferation and induce senes- the demand of differentiation therapy. Our previous cence (Fig. 6c, d). Thus, inhibition of UGTs led to study found that both NPC cell lines and NPC tissues differentiation of NPC cells. showed decreased expression of SOX1 at the mRNA and protein levels, which was associated with promoter Discussion hypermethylation. Ectopic expression of SOX1 repressed Our new data provided evidences to suggest a novel NPC cell proliferation, migration, and induced cell dif- therapeutic strategy for application of differentiation ferentiation16. However, the mechanism SOX1-induced

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 12 of 15

Fig. 6 (See legend on next page.)

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 13 of 15

(see figure on previous page) Fig. 6 Targeting UGTs promotes differentiation of NPC cells. a Morphology of wild type HONE1 and CNE2 cells with or without T-96 treatment for 72 h. Scale bar = 50 µm. b Western blot analysis of KRT5, KRT13, and β-actin of wild type HONE1 and CNE2 cells with 0, 1, 5, 10, 20 μM T-96 treatment for 72 h. β-actin was used as a loading control. c Cell viability of wild type HONE1 and CNE2 cells with (red) or without (blue) T-96 treatment by CCK-8 assay. All data represent the mean ± SD (n = 4, **P < 0.01, ***P < 0.001, ****P < 0.0001). d SA-β gal staining (left panel) of wild type HONE1 and CNE2 cells with or without doxycycline treatment for 7 days. Red arrows represent SA-β gal-positive cells. Scale bar = 50 μm. Dot plots (right panel) show quantification of the frequency of SA-β gal-positive cells in each vision. All data represent the mean ± SD (n = 4, ***P < 0.001, ****P < 0.0001). e Summary of the present study. Retinoic acid (RA) is balanced to low level concentration in SOX1 low NPC cells, while SOX1 induces RA accumulation within NPC cells, promoting cell differentiation (upper panel). For treatment strategy, targeting at UGT2B7 induces NPC cell differentiation, which is associated with RA metabolism. cell differentiation remains unclear. Here, we found that UGTs catalyze the transfer of glucuronic acid to lipophilic the deletion or mutation of either DNA-binding domain substrates, and convert them into hydrophilic compounds or transcription activation domain of SOX1 impaired the that can be excreted from cells. Crystal structure of differentiation of NPC cells (Fig. 2 and Supplementary UGT2B7 is recently identified, which catalyzes the cova- Fig. 2). Thus, SOX1 promoting differentiation depended lent addition of glucuronic acid sugar moieties to a host of on its transcriptional function. therapeutics and environmental toxins31. UGT2B7 has Besides, we found that a global keratins high expression also been showed as responsible for the glucuronidation event in NPC tissues compared to normal tissues (Fig. of ATRA with high catalytic efficiency30. The drug- 3a). Ectopic expression of SOX1 in NPC cells down- induced UGTs expression has been repeatedly observed regulated the expression of keratins, including KRT5, after exposure of cancer cells to anti-cancer drugs. KRT13, and KRT19, which was similar to the gene UGT2B7 expression was demonstrated in human mela- expression profiles between NPC tissues and normal nocytes, but not cell lines derived from metastatic mela- tissues (Fig. 3a–d). As keratins family proteins are nomas. Treatment of these cell lines with anti-cancer potential markers for defining differentiation status of agents, including vemurafenib, induced expression of NPC cells26,27, these results furtherly demonstrated UGT2B732. Epirubicin upregulated UGT2B7 expression SOX1-induced NPC cell differentiation. in hepatocellular carcinoma HepG2 and Huh7 cells, Moreover, we cultured cells with CM obtained from promoting its own detoxification via the p53-mediated cells with SOX1 overexpression. Cells treated with CM pathway33. Besides, a further study showed a similar showed decreased keratins, suggesting that components induction of UGT2B7 by several other cytotoxic drugs, in CM promoted differentiation of NPC cells (Fig. 4c). including three anthracyclines (doxorubicin, daunor- LC/MS is currently the most widely used method of ubicin, and idarubicin) and six nonanthracyclines (mito- determining metabolic phenotypes via both untargeted mycin C, 5-fluorouracil, camptothecin, 7-ethyl-10- and targeted analysis. Therefore, in order to determine hydroxycamptothecin, topotecan, and etoposide) in components that induced the differentiation of NPC cells, hepatocellular carcinoma cells34. Thus, drug-induced untargeted metabolic LC/MS for CM and cells were UGT2B7 activity in cancer cells affected the therapeutic performed, and high level of retinoids was detected both efficacy of cytotoxic drugs and noncytotoxic drugs, which in CM and cells with SOX1 overexpression (Fig. 4d, e). were UGT2B7 substrates. It was shown that ATRA- We and others reported that ATRA inhibited prolifera- induced transcription via a RARE in the presence of tion, and induced differentiation of NPC cells18,28. Our RARs35. Although glucuronidation of ATRA might pro- results showed that cells treated with RAce or RA pre- duce retinoyl β-glucuronide or other metabolites, retinoyl sented reduced keratins expression, declined proliferation β-glucuronide binds to neither RARs nor CRABPs36,37. and colony formation (Fig. 4f–h). Hence, retinoid meta- Thus, RA glucuronidation might result in attenuating the bolic pathway was involved in SOX1-induced cell differ- differentiation induced by SOX1 in NPC cells. entiation. Previously, it was reported that SOX2 could T-96 was reported to inhibit UGT1A6 (Ki = 0.6 μM) affect glycolysis by regulating GLUT1, therefore con- and UGT2B7 (Ki = 17.3 μM)38. It was originally used as tributing to the disease progression of squamous cell an immunosuppressive agent, which significantly inhib- carcinoma29. But it remains elusive whether SOX1 could ited the activation of NF-κB39. Later studies reported that regulate metabolic-related genes and pathways. it also had anti-cancer effects. T-96 inhibited cell pro- Genes in regulating retinoid metabolic pathway were liferation and induced cell apoptosis through down- screened. The expression level of UGTs was significantly regulating the expression of MCL1 in melanoma cells40. reduced in poorly differentiated cells after doxycycline T-96 inhibited triple-negative breast cancer invasion by induction (Fig. 5d). CYP26s or UGTs were reported to blocking the canonical and non-canonical TGF-β signal- clear RA by hydroxylation or glucuronidation21,22,30. ing pathways41. And it suppressed glioma growth by

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 14 of 15

42 regulating the miR-30e-5p/MYBL2 axis . These results 6. Wood,H.B.&Episkopou,V.Comparative expression of the mouse Sox1, Sox2 demonstrated that T-96 might act as a promising agent and Sox3 genes from pre-gastrulation to early somite stages. Mech. Dev. 86, 197–201 (1999). for the treatment of cancers. Here, we found T-96 pro- 7. Nishiguchi, S., Wood, H., Kondoh, H., Lovell-Badge, R. & Episkopou, V. Sox1 moted the differentiation of NPC cells (Fig. 6a, b). directly regulates the gamma-crystallin genes and is essential for lens devel- Moreover, T-96 significantly inhibited HONE1 and CNE2 opment in mice. Gene Dev. 12,776–781 (1998). 8. Wegner, M. All purpose Sox: the many roles of Sox proteins in gene cell proliferation and induced cell senescence (Fig. 6c, d). expression. Int. J. Biochem. Cell Biol. 42,381–390 (2010). Therefore, UGTs could be used as targets for differ- 9. Graham, V., Khudyakov, J., Ellis, P. & Pevny, L. SOX2 functions to maintain entiation therapy in NPC treatment. neural progenitor identity. Neuron 39,749–765 (2003). 10. Sarkar, A. & Hochedlinger, K. The sox family of transcription factors: versatile Collectively, we showed that SOX1 promoted NPC regulators of stem and progenitor cell fate. Cell Stem Cell 12,15–30 (2013). differentiation by enhancing retinoids accumulation 11. Sinclair, A. H. et al. A gene from the human sex-determining region encodes a through deregulating UGTs expression. Our data provide protein with homology to a conserved DNA-binding motif. Nature 346, 240–244 (1990). compelling biomolecular basis to study UGTs as targets 12. Herman, J. G. & Baylin, S. B. Gene silencing in cancer in association with for differentiation therapy and reveal the potential use of promoter hypermethylation. N. Engl. J. Med. 349, 2042–2054 (2003). its inhibitor T-96 for NPC treatment (Fig. 6e). 13. Hu, J. et al. Sex-determining region Y box-containing genes: regulators and biomarkers in gynecological cancers. Cancer Biol. Med. 16,462–474 (2019). 14. Song, L. et al. SOX1 inhibits breast cancer cell growth and invasion through Acknowledgements suppressing the Wnt/β-catenin signaling pathway. APMIS 124,547–555 (2016). We thank the members of Liuq lab at Sun Yat-sen University for their technical 15. Tsao, C. M. et al. SOX1 functions as a tumor suppressor by antagonizing the support. This study was supported by the National Key Research and WNT/beta-catenin signaling pathway in hepatocellular carcinoma. Hepatology Development Program of China (No. 2017YFA0505600-04 to Q.L.), the National 56,2277–2287 (2012). High Technology Research and Development Program of China (No. 16. Guan, Z. et al. SOX1 down-regulates beta-catenin and reverses malignant 2015AA020926 to Z.-J.L.), the National Natural Science Foundation of China phenotype in nasopharyngeal carcinoma. Mol. Cancer 13, 257 (2014). (No. 81630005 to Q.L.; No. 81572648 to Z.G.), the Guangdong Basic and 17. Lin, Y. W. et al. SOX1 suppresses cell growth and invasion in cervical cancer. Applied Basic Research Foundation (Nos. 2018A030313071, 2019A1515011185 Gynecol. Oncol. 131, 174–181 (2013). to Z.-J.L.; No. 2018A0303130299 to M.Y.), the Science and Technology Planning 18. Yan, M. et al. IKKalpha restoration via EZH2 suppression induces nasophar- Project of Guangzhou (No. 201804020044 to Q.L.) and the Fundamental yngeal carcinoma differentiation. Nat. Commun. 5, 3661 (2014). Research Funds for the Central Universities (No. 18ykzd01 to Z.-J.L.). 19. Rhinn, M. & Dolle, P. Retinoic acid signalling during development. Develop- ment 139,843–858 (2012). Author details 1 20. Marill, J., Idres, N., Capron, C. C., Nguyen, E. & Chabot, G. G. Retinoic acid Sun Yat-sen University Cancer Center, State Key Laboratory of Oncology in – 2 metabolism and mechanism of action: a review. Curr. Drug Metab. 4,110 South China, 510060 Guangzhou, China. Department of Otorhinolaryngology, (2003). Sun Yat-sen Memorial Hospital, Sun Yat-sen University, 510120 Guangzhou, 21. Isoherranen, N. & Zhong, G. Biochemical and physiological importance of the 3 fi China. Department of Hematology, The Third Af liated Hospital, Sun Yat-sen CYP26 retinoic acid hydroxylases. Pharmacol. Ther. 204, 107400 (2019). University, Institute of Hematology, Sun Yat-sen University, 510630 Guangzhou, 22. Bock, K. W. Roles of human UDP-glucuronosyltransferases in clearance and China homeostasis of endogenous substrates, and functional implications. Biochem. Pharmacol. 96,77–82 (2015). Conflict of interest 23. Little, J. M. & Radominska, A. Application of photoaffinity labeling with [11,12- The authors declare that they have no conflict of interest. 3H]all-trans-retinoic acid to characterization of rat liver microsomal UDP-glucuronosyltransferase(s) with activity toward retinoic acid. Biochem. Biophys. Res. Commun. 230,497–500 (1997). Publisher’s note 24. Little, J. M., Lehman, P. A., Nowell, S., Samokyszyn, V. & Radominska, A. Glu- Springer Nature remains neutral with regard to jurisdictional claims in curonidation of all-trans-retinoic acid and 5,6-epoxy-all-trans-retinoic acid: published maps and institutional affiliations. activation of rat liver microsomal UDP-glucuronosyltransferase activity by alamethicin. Drug Metab. Dispos. 25,5–11 (1997). Supplementary Information accompanies this paper at (https://doi.org/ 25. Lu,Y.,Bratton,S.,Heydel,J.M.&Radominska-Pandya,A.Effectofretinoidson 10.1038/s41419-020-2513-1). UDP-glucuronosyltransferase 2B7 mRNA expression in Caco-2 cells. Drug Metab. Pharmacokinet. 23,364–372 (2008). 26. Reddy,S.P.,Raslan,W.F.,Gooneratne,S.,Kathuria,S.&Marks,J.E.Prognostic Received: 27 October 2019 Revised: 8 April 2020 Accepted: 8 April 2020 significance of keratinization in nasopharyngeal carcinoma. Am.J.Otolaryngol. 16,103–108 (1995). 27. Barak, V., Goike, H., Panaretakis, K. W. & Einarsson, R. Clinical utility of cytoker- atinsastumormarkers.Clin. Biochem. 37,529–540 (2004). 28. Zhang, W. et al. SPLUNC1 is associated with nasopharyngeal carcinoma References prognosis and plays an important role in all-trans-retinoic acid-induced 1. Ferlay,J.etal.Cancerincidenceandmortalityworldwide:sources,methods growth inhibition and differentiation in nasopharyngeal cancer cells. FEBS J. and major patterns in GLOBOCAN 2012. Int. J. Cancer 136,E359–E386 (2014). 281, 4815–4829 (2014). 2. Chua,M.,Wee,J.,Hui,E.P.&Chan,A.Nasopharyngealcarcinoma.Lancet 387, 29. Hsieh, M. H. et al. p63 and SOX2 dictate glucose reliance and metabolic 1012–1024 (2016). vulnerabilities in squamous cell carcinomas. Cell Rep. 28,1860–1878 (2019). 3. Tallman, M. S. et al. All-trans-retinoic acid in acute promyelocytic leukemia. N. 30. Samokyszyn, V. M. et al. 4-hydroxyretinoic acid, a novel substrate for human Engl.J.Med.337,1021–1028 (1997). liver microsomal UDP-glucuronosyltransferase(s) and recombinant UGT2B7. J. 4. Matthay, K. K. et al. Long-term results for children with high-risk neuro- Biol. Chem. 275,6908–6914 (2000). blastoma treated on a randomized trial of myeloablative therapy followed by 31. Miley, M. J. et al. Crystal structure of the cofactor-binding domain of the 13-cis-retinoic acid: a children’s oncology group study. J. Clin. Oncol. 27, human phase II drug-metabolism enzyme UDP-glucuronosyltransferase 2B7. J. 1007–1013 (2009). Mol. Biol. 369,498–511 (2007). 5. Gubbay, J. et al. A gene mapping to the sex-determining region of the mouse 32. Dellinger, R. W., Matundan, H. H., Ahmed, A. S., Duong, P. H. & Meyskens, F. L. Y chromosome is a member of a novel family of embryonically expressed Anti-cancer drugs elicit re-expression of UDP-glucuronosyltransferases in genes. Nature 346, 245–250 (1990). melanoma cells. PLoS ONE 7, e47696 (2012).

Ofﬁcial journal of the Cell Death Differentiation Association Lei et al. Cell Death and Disease (2020) 11:331 Page 15 of 15

33. Hu, D. G., Rogers, A. & Mackenzie, P. I. Epirubicin upregulates UDP glucur- 38. Zhao,J.W.,Wang,G.H.,Chen,M.,Cheng,L.H.&Ji,X.Q.Demethylzeylasteral onosyltransferase 2B7 expression in liver cancer cells via the p53 pathway. Mol. exhibits strong inhibition towards UDP-glucuronosyltransferase (UGT) 1A6 and Pharmacol. 85,887–897 (2014). 2B7. Molecules 17,9469–9475 (2012). 34. Hu, D. G., Mackenzie, P. I., Lu, L., Meech, R. & McKinnon, R. A. Induction of 39. Hu, Q., Yang, C., Wang, Q., Zeng, H. & Qin, W. Demethylzeylasteral (T-96) human UDP-Glucuronosyltransferase 2B7 gene expression by cytotoxic treatment ameliorates mice lupus nephritis accompanied by inhibiting acti- anticancer drugs in liver cancer HepG2 cells. Drug Metab. Dispos. 43,660–668 vation of NF-kappaB pathway. PLoS ONE 10, e133724 (2015). (2015). 40. Zhao, Y. et al. Demethylzeylasteral inhibits cell proliferation and induces 35. Idres, N., Marill, J., Flexor, M. A. & Chabot, G. G. Activation of retinoic acid apoptosis through suppressing MCL1 in melanoma cells. Cell Death Dis. 8, receptor-dependent transcription by all-trans-retinoic acid metabolites and e3133 (2017). isomers. J. Biol. Chem. 277,31491–31498 (2002). 41. Li, L. et al. Demethylzeylasteral (T-96) inhibits triple-negative breast cancer 36. Mehta,R.G.,Barua,A.B.,Olson,J.A.&Moon,R.C.Retinoidglucuronidesdo invasion by blocking the canonical and non-canonical TGF-beta signaling not interact with retinoid binding proteins. Int. J. Vitam. Nutr. Res. 62,143–147 pathways. Naunyn Schmiedebergs Arch. Pharm. 392,593–603 (2019). (1992). 42. Zhang, K. et al. Demethylzeylasteral inhibits glioma growth by regulating the 37. Sani, B. P., Barua, A. B., Hill, D. L., Shih, T. W. & Olson, J. A. Retinoyl beta- miR-30e-5p/MYBL2 axis. Cell Death Dis. 9,1035(2018). glucuronide: lack of binding to receptor proteins of retinoic acid as related to biological activity. Biochem. Pharmacol. 43,919–922 (1992).

Ofﬁcial journal of the Cell Death Differentiation Association